A General Framework for Information Extraction using Dynamic Span Graphs
This work addresses the challenge of integrating coreference and relation extraction in a unified way for researchers and practitioners in NLP, though it is incremental as it builds on existing multi-task frameworks.
The authors tackled the problem of information extraction across multiple tasks by introducing a dynamic span graph framework that refines span representations through confidence propagation, achieving state-of-the-art performance with significant F1 score improvements on datasets like ACE.
We introduce a general framework for several information extraction tasks that share span representations using dynamically constructed span graphs. The graphs are constructed by selecting the most confident entity spans and linking these nodes with confidence-weighted relation types and coreferences. The dynamic span graph allows coreference and relation type confidences to propagate through the graph to iteratively refine the span representations. This is unlike previous multi-task frameworks for information extraction in which the only interaction between tasks is in the shared first-layer LSTM. Our framework significantly outperforms the state-of-the-art on multiple information extraction tasks across multiple datasets reflecting different domains. We further observe that the span enumeration approach is good at detecting nested span entities, with significant F1 score improvement on the ACE dataset.